Research on Underwater Noise Features Based on Spectrum Analysis and Welch Algorithm

To explore the characteristics of underwater noise, we analyze the power spectrum and frequency of underwater noise with the sequence diagram, the period diagram, the density diagram and improved welch algorithm according to the 5-day noise data by root mean square. The results show that, first of all, spectrum analysis and welch algorithm simplify the calculation process and adapt to the impulsiveness and randomness of noise. Secondly, adopting different window functions in welch power spectrum estimation improves the spectral resolution of underwater noise prediction, which is insensitive to noise. Thirdly, in the spectrum analysis, the accuracy of experimental results from welch algorithm obviously outperforms sequence diagram, periodic diagram and density diagram. Compared with periodic diagram and density diagram, the improved welch algorithm will obtain smoother power spectrum estimation, reducing the resolution and random error of noise frequency in the segmentation process. Finally, when the number of sites is greater than 10, our simulation results are more accurate than that with periodic diagram method.